Configure AI Agents

Video walkthrough

What this does

Agents are configurable AI assistants that live inside Bruin Cloud. Each agent can be scoped to a project, connected to messaging platforms, and given a connection set that controls exactly which data it can read. You can also let it run Bruin Cloud CLI commands and give it a custom system prompt.

Steps

1) Open the agents page

From the AI menu, go to Agents. You'll see the list of existing agents and a button to create a new one.

2) Pick a project

The first step is selecting the project the agent should connect to.

• Pick a project - the agent gets access to that project's repo and pipelines.
• No project - the agent has no access to your repos or projects. It behaves like a regular ChatGPT or Claude chat.

3) Name the agent

Give the agent a clear name. This is how it shows up in the agents list and in any messaging integrations.

4) Add messaging integrations

You can integrate the agent into messaging platforms like Slack, Discord, Teams, and WhatsApp.

• For Slack and Discord, set up the integration once from the Integrations menu. After that, when you create an agent you can pick the integration and enter the channel ID.
• For other platforms, you generate a code in Bruin Cloud and follow the on-screen instructions to wire it up.

If you don't pick an integration, the agent is still usable - it lives inside the Bruin Cloud web chat, the dashboard builder, and as a scheduled agent.

5) Attach a connection set

Under Connections, pick the connection set the agent should use.

A connection set is a named bundle of connections to data platforms, kept separate from the connections your pipelines use. This separation lets you:

• Restrict agents to only the data they need.
• Give agents read-only access where pipelines have read/write.
• Apply granular, agent-specific permissions without touching pipeline credentials.

You create connection sets from Connections settings - click New connection set, name it, and pick the data platform connections it should include.

If you create an agent with no connection set, it can still answer general questions and help with non-data tasks, but it won't be able to read your data. It behaves more like a regular ChatGPT or Claude.

6) Optional - Cloud CLI access

You can also give the agent access to the Bruin Cloud CLI. With this enabled, the agent can:

• Run pipelines and assets
• Read logs and run history
• Query the data catalog and glossary

7) Optional - System prompt

Add a system prompt to give the agent specific instructions, role, or constraints. This is useful when you want the agent to focus on a particular workflow or follow a specific tone.

8) Create the agent

Click Create. The agent is now available in the agents list.

9) Reconfigure later

Open an agent from the agents list to change its settings at any time:

• Add or remove integrations
• Swap the connection set
• Edit the system prompt
• Manage access - control which teams and members in your organization can use the agent

Key takeaways

• Agents can be project-scoped or general-purpose.
• Connection sets separate agent permissions from pipeline credentials.
• Messaging integrations are optional - agents always work in the Bruin Cloud chat.
• Cloud CLI access lets agents run pipelines and read catalog metadata.
• Everything is editable later from the agent's settings page.

Next

To let an agent reach your data warehouse, set up the connections it needs first - see Manage Connections.